Attempts to demonstrate relationships between neuroleptic plasma levels and antipsychotic response have met with mixed results. Eight studies have attempted to clarify such discrepancies by contrasting drug level-response relationships in which the fit of plasma neuroleptic concentrations with response could be contrasted to the fit with antipsychotic response found with free (unbound) neuroleptic (or close derivations of free drug, such as cerebrospinal fluid or erythrocyte neuroleptic concentrations). In describing such drug level-response relationships, seven of eight studies show superiority of free, cerebrospinal fluid or erythrocyte concentrations of drug, as compared with plasma neuroleptic levels (p less than 0.05). Variance between neuroleptic blood levels and therapeutic response was reduced 12.0% +/- 17.7% (SD) by the use of free, cerebrospinal fluid or erythrocyte neuroleptic concentrations rather than plasma drug concentrations. Neuroleptic assays that monitor metabolites as well as parent neuroleptic may also reduce variance between drug level and response. The state of present information is such that routine monitoring of neuroleptic drug concentrations to guide dosage adjustment is premature.